Currently, networks are developing rapidly. In any scenario, service providers strive to simplify the device, reduce device costs and device management costs, and improve the speed of service convergence in the case of failover.
In the prior art, a Next Generation Network (NGN) accesses a Provider Edge Router (PE) of the IP/MPLS network directly in an active-standby mode. The details about how an NGN device accesses the IP/MPLS network in an active-standby mode in the prior art are described below, taking a Multimedia Gateway (MGW) among NGN devices as an example.
FIG. 1 shows how an MGW accesses an IP/MPLS network in the prior art. As shown in FIG. 1, the MGW works in an active-standby mode, and is directly connected with two PEs (PE 1 and PE2) of the IP/MPLS network. The active port on the MGW is connected with PE1 through active link 1 and active link 2, and the standby port is connected with PE2 through standby link 3 and standby link 4. Link 3 is a standby link of link 1, and link 4 is a standby link of link 2. Each active link and its corresponding standby link have the same IP address. Normally, the standby port of the MGW does not work, namely, does not receive or send messages. Therefore, standby links do not receive or send data streams. The active port of the MGW sends Address Resolution Protocol (ARP) request messages to PE1 periodically through the active link, and PE1 returns ARP response messages after receiving the ARP request messages. If the MGW does not receive the ARP response message from PE1 within a preset time, the MGW determines that the active link fails and triggers active-standby failover. That is, the standby port changes to an active port, and the standby link changes to an active link.
Virtual Router Redundancy Protocol (VRRP) and Virtual Private LAN Segment (VPLS) are applied on PEs. According to the VRRP, PE1 is set as an active device, PE2 is set as a standby device, and the interface IP address of PE1 is set as the virtual IP address of the VRRP. A loopback board is set inside the two PEs. That is, the active PE1 runs the physical port of the VRRP, and sends a VRRP multicast message periodically. The VRRP multicast message is carried by a VPLS and flooded to the physical port that runs the VRRP on PE2. If the standby PE receives no VRRP multicast message within three sending periods for sending VRRP multicast messages, the standby PE determines that the active PE fails, and triggers VRRP active-standby failover and the standby PE changes to a active PE. PE1 and PE2 are located in the same subnet, and each advertises routes to a remote PE3, as shown by dotted lines in FIG. 1.
Normally, the MGW forwards traffic to PE1 through an active port and the active link connected with the active port, and then PE1 forwards the traffic to PE3 or another PE through an IP/MPLS network.
Part of the return traffic sent from PE3 to PE1 through a route advertised by PE1 is directly forwarded to the MGW through the active link, as shown by the bidirectional arrow in FIG. 1.
Part of the return traffic sent from PE3 to PE2 through a route advertised by PE2 is transmitted to PE1 transparently through a VPLS network between PE2 and PE1, and then forwarded to the MGW through the active link, because the MGW standby interface connected with PE2 is unable to receive or send traffic, as shown by the unidirectional arrow in FIG. 1.
When the MGW detects that an active link fails, for example, active link 1 fails, the MGW triggers active-standby failover. And, the active port connected with active link 1 changes to a standby port, the old active link changes to a standby link which will not receive or send data any more, the standby port connected with standby link 3 changes to an active port, and the old standby link changes to an active link which begins receiving and sending messages. In this case, PE1 and PE2 still work normally, and the VRRP does not trigger active-standby failover. FIG. 2 shows how an MGW accesses an IP/MPLS network when active link 1 fails in the prior art. As shown in FIG. 2, the MGW sends traffic to PE2 through the active port and link 3. Because the VRRP does not trigger active-standby failover after the active-standby failover of the MGW, and PE2 is still standby, PE2 transmits the received traffic to the L3VPN (Layer 3 Virtual Private Networks) port of PE1 transparently through the VPLS network via L3VPN port of PE2, and then PE1 forwards the traffic to PE3 or other PEs.
Part of the return traffic sent from PE3 to PE2 through the route advertised by PE2 is forwarded to the MGW through the active link after failover and the active port after failover, as shown by the bidirectional arrow in FIG. 2.
Part of the return traffic sent from PE3 to PE1 through the route advertised by PE1 is transmitted to PE1 transparently through the VPLS network between PE2 and PE1 and then forwarded to the MGW through the active link after failover and the active port because the old active port after failover has changed to a standby port which will not receive or send messages any more, as shown by the unidirectional arrow in FIG. 2.
As described above, in the prior art, the MGW sends ARP detection messages to PE1, and receives ARP response messages returned by PE1 to detect the fault of the active link, and detects fault of the PE through VRRP.
If the physical link between PE1 and PE2 fails, normally PE2 is unable to forward the traffic received from the remote PE to PE1; after active-standby failover of the MGW, PE1 is unable to forward the traffic received from the remote PE to PE2, thus causing serious loss of service packets.
When the active link fails, the MGW undergoes active-standby failover. However, because PE1 does not fail, neither PE1 nor PE2 undergoes active-standby failover. Therefore, the sent traffic and a part of the return traffic need to be forwarded through active PE1, which increases the traffic forwarding time and slows down the service convergence.
Moreover, in the prior art, when the network device accesses the IP/MPLS network through a PE, a loopback board needs to be configured on the PE in order to run the VRRP and the VPLS, and therefore, the PE is rather complicated. In order to improve the reliability of the VPLS, two physical links need to be configured between PE1 and PE2 to ensure transparent transmission of the traffic of the VPLS network. That increases the device costs, probability of faults of the whole system, and device management costs.
Such problems also occur on other network devices which access the packet switched network in an active-standby mode.